Circular correlation for uniform marginals about pingouin HOT 6 CLOSED

Hi @dominikstrb ! This is interesting, thanks for sharing! I, unfortunately, don't have access to the book, any chance that you could send me a screenshot of the relevant paragraph/equation?

Here are a few thoughts:

1. I would probably leave the current method as the default, to be consistent with the Matlab circular toolbox.

2. However, we can indeed add an optional argument for this method. What do you think is the most appropriate argument name? correction_uniform=True, method='js', or any other ideas?

3. Do you know of any packages / softwares (R, Matlab, Python, SPSS) that implement this equation? It would be better for testing purposes.

Thanks!
Raphael

from pingouin.

Thanks for your quick answer!

I took a screenshot of the relevant page:

My answers to your thoughts:

1. Definitely!
2. I think as long as there's only one alternative method, one boolean kwarg should be enough, if the user is referred to the relevant literature.
3. I wasn't able to find one, unfortunately. Most circular statistics packages (like this one) seem to build upon Philipp Berens' MATLAB toolbox. I'll make sure to test the two methods against one another with simulations before proposing any changes to Pingouin.

If you're okay with it, I could start a pull request that keeps the default method, but allows to enable the alternative with a kwarg.

Best,
Dominik

from pingouin.

Hi @dominikstrb ! Yes, please start working on a pull request :) Make sure to push on the develop, and not the master, branch. Thanks!

from pingouin.

Hi @dominikstrb! Just wanted to follow-up on the PR. Please let me know if you need any help, or, if you're too busy at the time I can also quickly add it.

Thanks!
Raphael

from pingouin.

Thanks for reminding me. I was in fact quite busy the past few weeks, but I managed to create the PR this morning. Hope everything is alright!

from pingouin.

Thanks for the PR @dominikstrb ! Closing the issue now.

from pingouin.